Welcome back to the second blog in our Ackron project case study. As we discussed in the first blog, traditional forest sampling methods are manual, time-consuming, and prone to human error. During technology tests at Ackron, a well-established forest planted in 2000, our work identified discrepancies between our numbers and those provided by the standard-approved Validating and Verification Bodies (VVBs). Therefore, post-survey work involved reconciling our numbers with those originally generated to understand why discrepancies arose. There were two main areas from which errors stemmed - tree number counts and measurements of the planted area - both of which combined led to a significant overestimate of aboveground carbon stocks or actual ‘ex-post’ carbon removal by verifiers. In this blog, we will be exploring how inaccuracies in the quantification of the project area arose, and how our technology improved measurement accuracy. 

In traditional sampling methods, the tree count estimated from a collection of sample plots is averaged and extrapolated to approximate tree count over the whole project site using the number of sample plot units. A sample plot unit is equivalent to the total project area divided by the sample plot area. It is important to note that woodland projects often have both gross site areas and net planted areas, with the net planted area being the actual land where trees are planted in a set project. Therefore, establishing an accurate measurement of the planted area (or project area) is important as over or underestimations can have ramifications for the accuracy of carbon stock calculations.

At Treeconomy, we use segmentation to separate the project area from a non-project area by constructing GIS vector polygons that represent the planting area. Segmentation allowed us to clearly delineate the project area by clipping boundaries to ensure only forested territory was measured, without including other lands such as fields, roads, water bodies, and bare ground. Segmentation has the added benefit of allowing us to optimise processing time and improve predictions of above-ground biomass - two identified drawbacks of traditional sampling techniques. 

At Ackron, you can see from the images below that the previous measurements used a crude and simple polygon to calculate the total planting area. Given that traditionally planting area forms a key metric in final carbon calculations, measurements the WCC generated from these polygons were critical building blocks in their final carbon stock calculation. In contrast to the crude and simple polygon the WCC and VVBs were relying on, the Treeconomy output looks very different. Segmentation of satellite imagery data allowed us to neatly and accurately partition off the planting zone, and gain an exact coverage of the project area in hectares - marking a significant improvement on the calculations made previously by the Woodland Carbon Code (WCC). 

In the initial project design, Ackron was listed with a gross area of 52.6 hectares (ha) and a project site of 46.98 ha. Through analysis, we determined the project area to be 40.1 ha - a 14.65% decrease versus the original classification. Replacing the original value in the WCC verification spreadsheet with our updated project area of 40.1 ha results in a significant reduction of total aboveground sequestration from 8,322tCO2 to 7,602tCO2 - an 8.7% decrease. 

These calculations demonstrate the enormous power of technology, AI, and data analytics in the quantifying project area and more accurately calculating ex-post carbon sequestration. As discussed, the crude nature of traditional sampling methods can, and does bring about errors in carbon stock estimation which has ramifications for verifiers, buyers, sellers, and the voluntary carbon market as a whole. Our post-survey work has illustrated how remote sensing technology and data analytics improve the precision with which project site areas can be measured, and thus improvements in the quantification of aboveground carbon stock. In contrast to traditional methods, our technology not only increases precision but also the speed and scalability of carbon stock quantification, which will be important as the voluntary carbon market and the number of carbon projects continue to grow. 

In our next blog, we will be digging into how technology can also bring consequential improvements to tree counting, and how by refining both project area and tree count, the carbon sequestration figures at Ackron were further refined.  To keep up to date with further announcements at Treeconomy HQ and how the initiatives are doing, make sure you follow us on our social media channels.